Edge Detection Algorithm for Color Noise-Laden Images in A Composite Color Space

Aiming at the non-uniform and unintuitive RGB color space, the HSI color space which is more in line with human color visual characteristics is used for edge detection. This paper proposes an improved multi-scale multi-structure mathematical morphology for edge detection of color noisy images. Firstly, the R, G, and B components of the color-containing noise image are extracted, and the R, G, and B components are separately subjected to morphological denoising preprocessing using structural elements of different scales. Then, the multi-scale top-hat transform is used to denoise. The image is contrast-enhanced; finally, the processed R, G, and B components are converted into H, S, and I components, and the H, S, and I components are edge-detected using the improved morphology, and the weight values are defined according to the image information entropy. Weighted fusion is performed to obtain the final color edge image. The experimental data shows that the edge detected by the method is more in line with human color visual characteristics, the denoising effect is good, and the contour and detail information of the original color image can be more retained.